Neuro-Fuzzy Classification: NEFCLASS 2.04 Update available

Detlef Nauck (
Mon, 5 Feb 1996 20:36:41 +0100


NEFCLASS-PC version 2.04 is now available.
This is a bug fix to version 2.0 that was released in October.

NEFCLASS-PC is software for classifying data by means of our neuro-fuzzy model
NEFCLASS (NEuro-Fuzzy CLASSification). NEFCLASS-PC runs on MS-DOS PC (80286 and
above), and it is freely available at (NEFCLASS homepage) or (480 KB zip file)

If you already have version 2.0 you can update by retrieving the zip file (190 KB)

NEFCLASS has been downloaded over 1900 times since version 1.5.

The following bugs are fixed:

* An error in the network file format was fixed.
* A problem in the specification for the pattern ranges was fixed.
* NEFCLASS did not include rules with a negative performance index
(i.e. rule activation for correct class is smaller than accumulated
activation for the other classes). Now it does, because this seems
to be necessary for complex problems.
* When a network was saved directly after creation, it could not be correctly
reloaded and trained. This was fixed.

If you cannot use WWW or FTP you can write me an e-mail to obtain the

If your are interested, please read the attached abstract from the
documentation of NEFCLASS-PC.

Best regards,

Detlef Nauck

Dr. Detlef Nauck Phone: 49.531.391.3155
Dept. of Computer Science Fax : 49.531.391.5936
Technical University of Braunschweig Email:
Bueltenweg 74 - 75 or:
D-38106 Braunschweig, Germany WWW:


An interactive simulation software to develop,
train, and test a Neuro-Fuzzy System for Classification

Version 2.04

February 1, 1996

1.) What ist NEFCLASS?

NEFCLASS is a neuro-fuzzy model based on a generic 3-layer fuzzy
perceptron, and it is used for classifying data (patterns).
NEFCLASS is trained with a set of patterns, where each pattern belongs
to one of a number of distinct classes (crisp
classification). NEFCLASS finds fuzzy rules by scanning the data, and
later optimizes these rules by learning the parameters of the fuzzy
sets that are used to partition the domain of the input variables
(features of the patterns).

After the learning process the resulting NEFCLASS system can be used
for classifying new, previously unknown data. The system can be
interpreted in form of fuzzy rules like

IF x1 is A1 and x2 is A2 and x3 is A3 and x4 is A4
THEN the pattern (x1,x2,x3,x4) belongs to class i,

where A1 - A4 are linguistic terms represented by fuzzy sets.

To obtain more information about the NEFCLASS model please read the
postscript file that is included in the distribution, or
check out our WWW pages at

2.) Where can I obtain NEFCLASS-PC?

NEFCLASS-PC may be freely used for educational, scientific or personal
purposes. NEFCLASS-PC is not free software, the copyright is held by
Technische Universitaet Braunschweig, Institut fuer Betriebssysteme
und Rechnerverbund (Technical University of Braunschweig, Dept. of
Computer Science). NEFCLASS-PC may not be used for commercial
purposes. Please read the licence file included in the distribution.

NEFCLASS-PC can be obtained free of charge by anonymous ftp at ( in the directory
/pub/local/nefcon, in the file (compressed with PKZIP 2.04g).

It can also be obtained via the World Wide Web at Please look for appropriate links on
this page.

If you are not able to use ftp or WWW, send an email, fax or letter to:

Dr. Detlef Nauck
Technical University of Braunschweig
Dept. of Computer Science
Bueltenweg 74-75
D-38106 Braunschweig
Tel: +49.531.391.3155
Fax: +49.531.391.5936
email: or

3. Whats New in Version 2.0?

* NEFCLASS-PC does now run in protected mode, and therefore needs an 80286
processor. It can make use of extended memory.
* A network can now have 50 input/output units, and up to 500 rule units.
* You can specify variable and class names when you create a new network.
* Each input variable can have an individual number of fuzzy sets.
* You can enter a description for a network.
* You can manually enter, delete and modifiy rules represented within a
* A network can be changed after it was created or trained.
* If a new network is specified, an eventually loaded pattern set will
be deleted from memory.
* The number of patterns that can be loaded does only depend on the
available memory.
* A pattern set can have a description (name).
* The learning parameters have been split in learning control parameters
and learning parameters.
* There are two new constraints for learning membership functions
(no passing, asymmetrical learning).
* All 3 rule learning procedures do now work.
* Rule learning is now independent from the sequence of patterns.
* The rule base can be relearned.
* A new dialog computes statistics of the pattern set, and can graphically
display bar charts of a feature, and 2-dimensional projections of the
patterns. These graphics can be plotted to an HPGL file.
* An entry was added to the Help menu that informs about the memory
* There is now a warning before leaving the program, if a network
was specified.
* NEFCLASS-PC looks for the environment variable BGIPATH to find the
graphic drivers (.bgi and .chr files). If this variable is not set, it
expects these files to be in the current directory.
* The online help was extended.